The Harsh Truth: AI Won’t Make Your Projects More Productive (Yet)

A contrarian view perhaps, but I like to challenge, provoke thought, and make people think differently. Given that PMI has just made its AI model available, I’m possibly in a small group of one who believes AI isn’t going to shift the productivity dial on projects just yet. Don’t get me wrong—I’m a HUGE fan of AI and believe it has the potential to transform how we deliver projects. But today, I don’t think it can live up to the promise and the hype.

What AI Can Do for Your Projects

Let’s look at some of the obvious things AI can do for our projects:

• Accelerate research time. Whether it’s reading reports (AI summaries) or asking questions (AI-generated answers).

• Requirements management. AI can create requirements, test them for ambiguities and contradictions, and even develop acceptance criteria.

• Email summarisation. AI can summarise emails and highlight the important ones, meaning I don’t have to be a constant hawk over my inbox.

• PowerPoint generation. AI can turn text prompts into polished presentations, which is great for those of us who aren’t the most artistic.

• AI Agents are on the horizon. These could soon handle tedious tasks like updating logs or running reports.

So Why Doesn’t This Make Us More Productive?

Firstly, we still have a privacy and security problem. We find it hard to share our dirty laundry internally in our organisations in a way that makes us feel safe. The willingness to potentially put that dirty laundry out into a third party domain, with all the privacy and security concerns that go with it, is just too big a leap for most people.

Second, AI models still hallucinate and produce inconsistencies that force me to redo the task myself. I’ve seen Copilot summaries of meetings that didn’t reflect what happened, and inbox summaries that missed crucial one-line emails, like “call me” from a sponsor. If I have to redo the work, no time is saved.

Third, we’re not experts at driving AI models yet. Try generating a governance deck using only prompts—it often takes as much time and effort to get the right output as it would to create the deck manually. There’s still a gap in either the models interpreting our requirements, or our expertise in defining requirements in a way that the models will give us the outputs we need. This gap is a big productivity sink today.

All 3 of these will erode the productivity gain - and they’re not even the big reason why projects and AI are not comfortable bedfellows just yet.

The Value Is in the Thinking

The value in producing the governance deck, the requirements, the research paper, is in the thinking that the production necessitates. Having to think about what is really important, having to think about what things can the governance committee action to improve, have to think about whether requirement X is really important to us, having to think about which emails I need to respond to is where the value is created. The value is created in the thinking, not in the production. If an AI system is able to produce, without us employing the same level of thinking, that erodes the value. If we move into a world where production is accelerated, but thinking is diminished, then understanding and meaningful action reduces too. To me, that doesn’t feel like it’s going to be a win for my project. AI might help me move at pace, but if I’m moving at pace in the wrong direction, that’s not going to be a productivity gain in the long run.

While I think AI will have it’s day and change how we do projects, I am fearful that we may move into an era of ‘production without understanding’ that will give us the illusion of progress, the illusion of more productivity, but miss the nuance of human relationships which is often the cause of most things that impact projects today. The micro interactions I have with people throughout the day are what help me as a Project Manager understand where I need to focus. The small pause in a question answered, the slightly less cheery response from a colleague, the lack of response - these are all the social clues that tell me what’s going on with the most important component of any project - the people. AI is not (yet) able to replicate human performance in picking up these signals and using them to influence project outcomes.

I think AI will have it’s day. But I also think we’re 2 steps into a thousand mile journey and there’s a long way to go before the productivity gains are real. I’m genuinely open to being wrong on this and I’m hopeful to get some interesting comments on my point of view. Let me know what you think.